The Los Alamos National Laboratory (LANL) has recently introduced the Venado supercomputer, a powerful tool designed for AI workloads in energy research and national security. With a performance capacity of ten exaFLOPS, Venado is a result of a collaboration between LANL, Nvidia, and Hewlett Packard Enterprise (HPE).

Located in New Mexico, the Venado supercomputer boasts 2560 Nvidia Grace Hopper Superchips and HPE's Slingshot 11 Network connectivity, specifically designed for AI computational requirements. The system is entirely liquid-cooled and features 920 Grace-Grace CPUs and 2560 GH200 superchips, totaling 3480 Nvidia Superchips. The GH200 is the primary system module for processing AI tasks, with a 72-core Grace CPU and 480GB of memory.

Thom Mason, LANL's director, stated that Venado would enhance national security and basic research by integrating artificial intelligence and machine learning. Early tests have shown that Venado can produce significant results in simulations for astrophysics and material science, with a focus on workflow efficiency rather than extreme accuracy.

The Venado supercomputer is a suitable candidate for large language models processing, as it prioritizes processing speed over accuracy. Although it falls short of the AMD Frontier system's capabilities, it can manage lower precision workloads, which are in high demand due to artificial intelligence requirements.

David Turk, the deputy secretary of the U.S. Department of Energy, expressed his excitement about Venado's ability to incorporate artificial intelligence approaches and deliver new and meaningful results for areas of interest.

Venado is a result of codesign by different vendors and a significant investment by the U.S. government, particularly under the Biden administration. With a focus on AI and machine learning research and development, Venado is a powerful tool for national security and energy research. However, the British government's project Isambard, which will house 5448 Nvidia GH200 Superchips, is expected to debut this year, potentially surpassing Venado's capabilities.